Journal of the Egyptian National Cancer Institute (2011) 23, 123–131

Cairo University

Journal of the Egyptian National Cancer Institute

www.nci.cu.adu.egwww.sciencedirect.com

ORIGINAL ARTICLE

Predictive value of immunohistochemical expression

of claudin-1 in colonic carcinoma

Eman Abdelzaher a,*, Azza Mohamed Rizk a, Samer Saad Bessa b,

Khalafalla Mustafa Omer c

a Pathology Department, Faculty of Medicine, Alexandria University, Alexandria, Egyptb Surgery Department, Faculty of Medicine, Alexandria University, Alexandria, Egyptc Ministry of Health, Sudan

Received 21 September 2011; accepted 8 October 2011Available online 9 November 2011

*

Eg

E-

11

Pr

Pe

do

Op

KEYWORDS

Colonic carcinoma;

Claudin-1;

Tight junctions;

Prediction of stage and

lymph node status

Corresponding author. Addr

ypt. Tel.: +20 101236757.

mail address: pathologistema

10-0362 ª 2011 National

oduction and hosting by Els

er review under responsibility

i:10.1016/j.jnci.2011.10.002

Production and h

en access under CC BY-NC-ND li

ess: 10 El

n@yaho

Cancer

evier B.V

of Cairo

osting by E

cense.

Abstract Objective: Colonic carcinoma is one of the most common cancers worldwide. Recently,

the possible involvement of claudin-1, one of the major tight junction proteins, in the process of

tumorigenesis has been suggested. Also, claudin-1 has emerged as a potential prognostic factor

in different types of tumors. The aim of this study was to detect caludin-1 expression in colonic car-

cinoma and to correlate its expression with clinicopathological variables in an attempt to delineate

its role as a potential new prognostic marker.

Material and methods: Immunohistochemical expression of claudin-1 was assessed in 50 Egyptian

patients with colonic adenocarcinoma. The predictive performance of claudin-1 expression was sta-

tistically evaluated.

Results: Decreased claudin-1 expression was found in 62% of colonic adenocarcinoma cases while

similar expression was found in 38%. Statistical analysis showed a statistically significant inverse

correlation between claudin-1 expression and tumor grade, depth of invasion, lymph node involve-

ment, and tumor stage. Regression analysis showed that claudin-1 decreased expression signifi-

cantly predicts that the tumor is of a high grade, high stage, and is associated with lymph node

galaa St. Victoria, Alexandria,

o.com (E. Abdelzaher).

Institute, Cairo University.

.

University.

lsevier

124 E. Abdelzaher et al.

involvement. ROC curve analysis showed that claudin-1 had a sensitivity of 88.24% and a specific-

ity of 81.25% for the prediction of tumor stage and a sensitivity of 73.33% and a specificity of

82.86% for the prediction of lymph node involvement.

Conclusions: Claudin-1 decreased expression in colonic carcinoma contributes to tumor dedifferen-

tiation, invasion and metastasis. Claudin-1 expression could be used as a predictor of colonic car-

cinoma stage and lymph node status with a high sensitivity and specificity.

ª 2011 National Cancer Institute, Cairo University. Production and hosting by Elsevier B.V.

Open access under CC BY-NC-ND license.

Introduction

Colonic carcinoma is the fourth most common cancer world-wide with an estimated one million new cases and a half milliondeaths each year [1]. In Africa colonic carcinoma accounts for

just 2.5% of all cancers compared to 9.5% in developed coun-tries [2]. According to the Egyptian National Cancer Institute,colonic carcinoma in Egypt contributes 6.5% of all cancers [3].

It is now clear that colonic carcinoma results from a multi-step accumulation of genetic alterations in proto-oncogenes,tumor suppressor genes and DNA repair genes [4]. In sporadiccolonic carcinoma, these alterations are acquired, and are

likely to be caused by exogenous and endogenous carcinogens.In contrast, in cancer syndromes, critical genetic alterationsthat predispose to malignancy are inherited [5].

Due to the increasing incidence of colonic carcinoma, sev-eral studies have been done in an attempt to understand itspathogenesis and assess the most valuable prognostic factor

[6]. Pathologic staging of colonic carcinoma is the best avail-able clinicopathologic prognostic marker, however, there isan urgent need to find markers that can stratify patients better

and earlier according to their risk of colonic carcinoma recur-rence and overall survival [7]. In recent years a large number ofprognostic markers have been suggested for colonic carcinoma[8–10].

Evidence for altered tight junctions in carcinomas has beenknown for more than 30 years [11]. Observations in examinedtumors include attenuation or lack of tight junctions, increased

paracellular permeability, and reductions in the number oftight junction strands which all result in a decrease in the epi-thelial barrier function of cells. Therefore, the loss of tight

junctions appears to constitute an essential step in cancerdevelopment [11–14]. Tight junction proteins also play criticalroles in cellular proliferation and neoplastic pathways via theirfunctions as transmitters of the extracellular environment to

intracellular signaling pathways [15].Claudins are a large family of essential tight junction pro-

teins [16]. There are two major functions defined for claudins:

the regulation of paracellular permeability and maintenance ofthe cell polarity [16].

The impact of claudin dysregulation deserves attention, as

it is now clear that these proteins play a role in tumorigenesisand represent promising new targets for cancer diagnosis,prognosis and therapy [17,18].

A multitude of studies reported down-regulation and up-regulation of various claudins in various cancers [19]. How-ever, only little information is available on the influence ofclaudin expression on tumor behavior. Also, relatively few

studies have examined the role of claudins in the tumorigenesisand prognosis of colonic carcinoma [20].

So far, 24 members of the claudin family have been identi-fied [21]. Claudin-1 which is ubiquitously expressed in many

organs, is thought to be a universal marker of tight junctionsand is alleged to be associated with the development of variouscancers [22]. However, the significance of claudin-1 expression

in cancer cells is not well understood [23]. In colonic carci-noma, the molecular and morphological alterations of clau-din-1 are still poorly understood [24–26].

The aim of this study was to detect caludin-1 expression in

colonic carcinoma and to correlate its expression with clinico-pathological variables in an attempt to delineate its role as apotential new prognostic marker.

Material and methods

Patients and tissue samples

The present work included 50 cases of colonic carcinomas thatwere retrieved from the surgical pathology archives at thePathology Department, Faculty of Medicine, the Alexandria

University between the years 2004 and 2009. Complete clinicaldata were available for all patients. None of these patients re-ceived pre-operative adjuvant chemotherapy or radiotherapy.

Pathological examination

Routinely processed paraffin-embedded tissues were cut into

5 lm-thick sections and stained with the conventional H&Estain. Each case was carefully reviewed and graded accordingto the WHO classification system of gastrointestinal tumors

[27]. In addition, the stage of the tumor was established accord-ing to the AJCC staging system [28].

Immunohistochemistry for claudin-1

Tissue sections (5 lm) were deparaffinized in xylene and rehy-drated through graded ethanol to water. Hydrogen peroxide

was applied to block endogenous peroxidase activity. Antigenretrieval was performed by boiling tissue sections in 10 mMcitrate buffer, pH 6.0. Slides were then incubated with the pri-

mary antibody (claudin-1, rabbit polyclonal antibody, ready touse). The bound antibody was detected by the UltraVisionDetection System (Anti-Polyvalent, HRP/DAB, ready-

to-use). Both the primary antibody and the detection kit werepurchased from the Lab Vision Corporation (Neo Markers,Fremont, USA). Immunohistochemical staining was per-formed using an avidin–biotinylated immunoperoxidase meth-

odology. Both positive and negative controls were included inall runs.

Predictive value of immunohistochemical expression of claudin-1 in colonic carcinoma 125

Assessment of Immunostaining

Acasewas considered claudin-1 positive if any number of tumorcells showed membranous and/or cytoplasmic positivity [19].

Both the intensity and the extent of claudin-1 immunostainingwere evaluated in tumor tissue as well as in the adjacent normalmucosa which served as an internal control.

The intensity of staining was graded as either weak = 1;moderate = 2; or strong = 3. The extent of staining was evalu-ated semiquantitatively and categorized as focal (610%) = 1;regional (11–50%) = 2; or diffuse (P50%) = 3.

Total scores for caludin-1 expression were then calculatedby multiplying the obtained scores of both intensity and extentof staining [19].

Total scores of claudin-1 expression in the studied coloniccarcinoma cases were compared to the adjacent normal muco-sa and a final computed score for each case was calculated (by

subtraction of the total score in normal mucosa from the totalscore in the tumor) [19].

In cases where total scores for claudin-1 expression in the tu-

morwere equal to the adjacent normalmucosa (final score = 0),the tumor was designated as having similar claudin-1 expressionto normal mucosa.

If the total score for claudin-1 expression in the tumor was less

than that in the normal mucosa (final score less than 0), the tumorwas designated as having decreased claudin-1 expression.

If the total claudin-1 staining score in the tumor was higher

than the normal mucosa; (final score more than 0), the tumorwas designated as having increased claudin-1 expression.

Statistical analysis

Datawere analyzed using PredictiveAnalytics Software (PASWStatistics 18). Significance of the obtained results was judged at

the 5% level.Non-parametric tests such as the Wilcoxon Mann–Whitney

test, Kruskal–Wallis test, and Spearman’s rho correlation were

used.Regression analysis was done. Linear models were devel-

oped aiming to identify the independent variable while control-

ling other variables.The predictive performance of claudin-1 expression in colo-

nic carcinoma was evaluated using Receiver Operating Char-

acteristic (ROC) curve analysis. The cutoff point at whichthe highest predictive accuracy was reached was estimatedusing Youden Index.

Results

Clinicopathologic study

The age of the patients ranged from 17 to 75 years (mean: 52.6,

median: 53.5). 18 patients (36%) were males and 32 patients(64%) were females with a M:F ratio of 1:1.8.

The most commonly involved part by the tumor was the ce-

cum (12 cases, 24%) followed by the ascending colon (10 cases,20%), the sigmoid colon (nine cases, 18%), the descending co-lon (seven cases, 14%), the transverse colon (six cases, 12%),

and the rectosigmoid area (six cases, 12%).Microscopically, all cases were classified as adenocarci-

noma (NOS). Histological grading was done according tothe WHO grading system. Twenty-three cases (46%) were well

differentiated, 18 cases (36%) were moderately differentiated

and nine cases (18%) were poorly differentiated.The 50 cases included in this study were staged according to

AJCC staging system. The tumor invasion was limited to thesubmucosa (T1) in only two cases (4%). Muscularis propria

invasion (T2) was noted in 15 cases (30%). In 31 cases(62%) the tumor reached the subserosa (T3) while in two cases(4%) the tumor directly invaded nearby organs or structures

(T4).As regards lymph node metastasis, 33 cases (66%) were des-

ignated as N0 while seven cases (14%) were designated as N1

and 10 cases (20%) were designated as N2.A total of 48 cases (96%) showed no distant metastasis

(M0), while only two cases (4%) showed distant metastasis

(M1) to the liver.Finally, stage grouping was done and 14 cases (28%) were

stage I, 20 cases (40%) were stage II, 14 cases (28%)were stageIII, and two cases (4%) were stage IV.

Immunohistochemical study

Immunoreactivity was interpreted without prior knowledge ofany of the clinicopathologic parameters. Both the tumor andadjacent normal colonic mucosa were separately scored for

claudin-1 expression.The normal colonic mucosa showed diffuse claudin-1 stain-

ing with a strong intensity in all the studied cases. The stainingpattern was membranous with accentuation of the lateral rather

than apical borders. Cytoplasmic staining was also noted. Thetotal score for normal mucosa was nine in all of the studiedcases. No staining was observed in the stroma (Fig. 1a).

Claudin-1 expression in colonic adenocarcinomas (Fig. 1b–f)was assessed regarding staining extent and intensity. The extentof staining ranged from focal (eight cases, 16%) to regional (19

cases, 38%) to diffuse (23 cases, 46%). Staining intensity wasstrong in 30 cases (60%),moderate in nine cases (18%) andweakin 11 cases (22%). Total scores ranged from 1 to nine with a

mean of 5.84 ± 2.999.The staining pattern was circumferential membranous and/

or cytoplasmic in well and moderately differentiated tumors.An incomplete membranous pattern was focally noted in mod-

erately differentiated tumors. In poorly differentiated tumors,the staining pattern was mainly cytoplasmic with minimalincomplete membranous staining.

Claudin-1 expression in the studied colonic adenocarcinomacaseswas similar to the normal colonicmucosa in 19 cases (38%)and showed decreased expression in 31 cases (62%)where the fi-

nal scores ranged from �3 to �8 (mean = �5.0968, med-ian = �5). None of the studied colonic adenocarcinoma casesshowed increased claudin-1 expression.

Table 1 illustrates the distribution of claudin-1 expressionregarding tumor grade, depth of invasion, lymph node involve-ment and tumor stage.

Relation between claudin-1 expression and clinicopathological

parameters (Table 2): statistical analysis showed that the relation be-tween claudin-1 expression onone hand and age, sex and tumor loca-tion on the other hand was insignificant (Spearman’s rho=�0.093,p=0.522, Mann–Whitney U=201.000, p=0.067, X2 = 8.490,p=0.131 respectively).

Astatistically significant strong inverse correlationwasdetected

between claudin-1 expressionononehandand tumorgrade, lymphnode status, and tumor stage in the other hand (Spearman’

Figure 1 Immunohistochemical expression of claudin-1. (a) Normal colonic mucosa showing diffuse membranous and cytoplasmic

claudin-1immunosatining of strong intensity (·100). (b) A case of well differentiated colonic adenocarcinoma showing similar claudin-1

expression (·40). (c) Higher power view of the previous case showing strong and diffuse circumferential membranous and cytoplasmic

claudin-1 immunostaining (·200). (d) A case of moderately differentiated colonic adenocarcinoma showing circumferential membranous

and cytoplasmic claudin-1 immunostaining. Incomplete membranous staining is noted (·400). (e) A case of poorly differentiated colonic

adenocarcinoma showing decreased claudin-1 expression (·40). (f) Higher power view of the previous case showing focal cytoplasmic

claudin-1 immunostaining of weak to moderate intensity (·400).

126 E. Abdelzaher et al.

rho= �0.676, p< 0.000, Spearman’s rho= �0.592, p< 0.000,and Spearman’s rho =�0.640, p< 0.000 respectively). In addi-

tion, a statistically significant, yet weak, inverse correlation wasfound between claudin-1 expression and depth of tumor invasion(Spearman’s rho= �0.280, p= 0.049) (Fig. 2).

Final score for claudin-1 expression in the two colonic carci-noma cases associated with metastasis (mean=�8, med-ian =�8) was lower than the final claudin-1 score in cases not

associated with metastasis (mean =�2.958, median =�3).However, statistical analysis could not be done due to small num-ber of the cases.

Regression analysis and ROC curve analysisThe clinicopathological factors found to be significantly corre-

lated with claudin-1 expression (the tumor grade, stage, depthof tumor invasion and lymph node status) were subjected tofurther regression statistical analysis. Three regression modelswere generated to overcome the problem of multicollinearity

(Table 3).From the developed models, claudin-1 decreased expression

was found to be a significant independent predictor for tumor

grade, stage and lymph node status. Decreased claudin-1 expres-sion significantly predicts that the tumor is of a high grade, high

stage, and is associatedwith lymphnode involvement (p< 0.001,p< 0.001, and p= 0.013 respectively).

A step further in analysis, ROC curve analysis was carried

out in order to detect a cut off value for claudin-1 expressionat which the highest predictive accuracy was reached. Thegrade was omitted from the analysis since tumor grade can

be easily assessed histologically by light microscopic exami-nation. A cutoff point was determined at final score �5(Fig. 3).

At the cutoff point of �5, claudin-1 showed a sensitivity of

88.24% and a specificity of 81.25% for the prediction of tumorstage. The positive likelihood ratio was 4.71 and the negative like-lihood ratio was 0.14. Predictive values were calculated with a po-

sitive predictive value of 90.9% and a negative predictive value of76.5%. The predictive power was good (area under thecurve = 0.892, 95% CI= 0.772–0.962, p=<0.000).

As regards lymph node involvement, at the cutoff point of�5, claudin-1 showed a sensitivity of 73.33% and a specificityof 82.86% for the prediction of lymph node status. The positive

Table 1 Distribution of claudin-1 expression.

Claudin-1 Similar expression Decreased expression

Number (%) Number (%)

Grade of the tumor WD 14 58.33 10 41.67

MD 5 29.41 12 70.59

PD 0 0 9 100

Tumor depth T1 0 0 2 100

T2 9 60 6 40

T3 9 29.03 22 70.97

T4 0 0 2 100

Lymph node involvement N0 18 54.55 15 45.45

N1 1 14.29 6 85.71

N2 0 0 10 100

Stage of the tumor I 10 71.43 4 28.57

II 9 45 11 55

III 0 0 12 100

IV 0 0 2 100

Table 2 Relation between claudin-1 expression and clinicopathologic parameters.

Claudin-1 final score Min Max Mean SD Median Test p Value

Age <40 �7.00 0.00 �3.600 2.608 �3.000 �0.093a 0.522

40� �8.00 0.00 �2.077 3.095 0.000

50� �8.00 0.00 �3.375 3.202 �3.00060� �8.00 0.00 �4.250 2.864 �3.00070–75 �5.00 0.00 �2.000 2.450 �1.500

Sex Male �8.00 0.00 �4.167 2.936 �4.000 Ub = 201.000 0.067

Female �8.00 0.00 �2.594 2.928 �3.000Tumor site Cecum �8.00 0.00 �2.250 3.166 0.000 (X2)c 8.490 0.131

Ascending colon �8.00 0.00 �5.300 2.627 �6.000Transverse colon �7.00 0.00 �3.333 3.141 �3.000Descending colon �8.00 0.00 �3.143 3.579 �3.000Sigmoid colon �6.00 0.00 �1.667 2.180 0.000

Rectosigmoid �7.00 0.00 �3.500 2.345 �3.000Grade of the tumor WD �5.00 0.00 �1.261 1.657 0.000 �0.676a <0.000*

MD �8.00 0.00 �3.611 2.790 �3.000PD �8.00 �3.00 �7.111 1.616 �8.000

Tumor depth T1 �7.00 0.00 �3.500 4.950 �3.500 �0.280a 0.049*

T2 �8.00 0.00 �2.200 3.075 0.000

T3 �8.00 0.00 �3.290 2.723 �3.000T4 �8.00 �8.00 �8.000 0.000 �8.000

LN involvement N0 �8.00 0.00 �2.057 2.496 0.000 �0.592a <0.000*

N1 �6.00 0.00 �3.800 2.388 �5.000N2 �8.00 �3.00 �6.700 2.003 �7.500

Metastasis M0 �8.00 0.00 �2.958 2.887 �3.000 NAd NAd

M1 �8. 00 �8.00 �8.000 0.000 �8.000Stage of the tumor I �7.00 0.00 �1.429 2.593 0.000 �0.640a <0.000*

II �8.00 0.00 �2.100 2.359 �3.000III �8.00 �3.00 �5.714 1.857 �5.500IV �8.00 �8.00 �8.000 0.000 �8.000

* Significant value.a Spearman’s rho correlation.b Mann–Whitney test.c Chi-square for Kruskal Wallis test.d not applicable.

Predictive value of immunohistochemical expression of claudin-1 in colonic carcinoma 127

likelihood ratio was 4.28 and the negative likelihood ratio was0.32. Predictive values were calculated with a positive predictive

value of 64.7% and a negative predictive value of 87.9%. Thepredictive power was good (area under the curve = 0.839,95% CI = 0.708 to 0.927, p= <0.000).

Discussion

Worldwide, every year more than one million individuals willdevelop colonic carcinoma, and disease specific mortality rate

is nearly 33% in the developed world [29].

a b

c d

Figure 2 Scatter plot graphs: (a) a strong inverse correlation between claudin-1 expression and tumor grade. (b) A weak inverse correlation

between claudin-1 expression and tumor depth. (c) A strong inverse correlation between claudin-1 expression and lymph node involvement. (d) A

strong inverse correlation between claudin-1 expression and tumor stage.

128 E. Abdelzaher et al.

Advances in molecular carcinogenesis, prognostic and pre-dictive molecular markers have been recently developed[1,30,31]. Among these markers are markers for tight junctions

[32]. The role of adherent junction proteins has been studiedextensively in cancer, but the roles of tight junction proteinsare less well understood [33]. Disruption of the cell–cell junc-

tion with concomitant changes in the expression of junctionalproteins is central to the cellular transformation and acquisi-tion of metastatic potential [33].

Claudins are integral to the structure and function of tightjunctions [33]. Recent studies show changes in expression/cel-lular localization of claudins during tumorigenesis; however,a causal relationship between claudin expression/localization

and cancer has not been definitely established [26].In the present study, 50 cases of colonic adenocarcinomas

were immunohistochemically studied for claudin-1 expression.

The nearby normal colonic mucosa was used as an internalcontrol and was also assessed for claudin-1 expression forcomparison. The extent, intensity and pattern of claudin-1

staining in normal colonic mucosa were similar to the results

reported by other investigators [15,19] who reported that theclaudin-1 staining was cytoplasmic with accentuation of thelateral cell membrane. In contrast Resnick et al. [15] reported

a circumferential claudin-1 immunostaining in the normal co-lonic mucosa.

As regards the studied colonic adenocarcinoma cases, a

greater number of cases (31 cases, 62%) showed decreasedclaudin-1 expression as compared to 19 cases (38%) thatshowed similar claudin-1 expression to the normal colonic mu-

cosa. This result is in line with several previous studies re-ported by Hahn-Stromberg et al. [34] and Miwa et al. [35].These authors reported a decreased claudin-1 expression in co-lonic adenocarcinomas. Also Tokes et al. [36] reported a de-

creased claudin-1 expression in invasive breast carcinoma ascompared to the normal breast ducts.

The mechanisms underlying the decreased claudin-1 expres-

sion in colonic carcinoma remain unknown. It is not quite asclear how down-regulation of claudins may contribute to theneoplastic progression [20]. However, it has been proposed

that decreased expression of claudin-1 with the loss of cell

a b

Figure 3 ROC curve: (a) showing the predictive performance of claudin-1 for stage. (b) Showing the predictive performance of claudin-1

for lymph node status.

Table 3 Regression models.

Regression models Variable Test p Value VIFa

Preliminary model Grade F= 24.987 <0.001

Stage t= �2.803 0.008 4.093

Lymph node involvement t= 0.845 0.403 3.979

Tumor depth t= �1.257 0.216 1.126

(Constant) t= 2.282 0.027

R2 = 0.625. F= 17.363, p < 0.001

Model 1 Grade F= 24.987 0.000

Stage �4.724 0.000

(Constant) 2.060 0.045

R2 = 0.652. F= 31.650, p < 0.001

Model 2 Grade F= 24.987 <0.001

Lymph node involvement t= �2.730 0.009

(Constant) t= �2.597 0.013

R2 = 0.556, F= 21.428, p < 0.001

a VIF: variation inflation factor.

Predictive value of immunohistochemical expression of claudin-1 in colonic carcinoma 129

polarity is followed by an abnormal influx of different growthfactors, which give rise to auto- and paracrine stimulations ofneoplastic epithelia and thus provides nutrition factors and

other factors necessary for tumor cell growth [37,38]. Also, itwas proved that the injured structure and function of tightjunctions might disorder the cell proliferation and differentia-tion and thus encourage the carcinogenesis [39].

A statistically significant strong inverse correlation wasfound between claudin-1 expression and the tumors’ grade inthe current study. These findings are in accordance with those

reported by Matsuoka et al. [40], Tzelepi et al. [41] andResnick et al. [15] who reported that decreased expression ofclaudin-1 is significantly associated with higher tumor grade

and poor differentiation. Moreover, Cha et al. [42] suggestedthat claudin-1 expression may be correlated with the progres-sion of colonic carcinoma.

In addition, Higashi et al. [22] reported that attenuatedexpression of claudin-1 closely correlates with the dedifferen-tiation of hepatocellular carcinoma and concluded that

down-regulated claudin-1 expression may thus serve as a po-tential marker for a poor prognosis in hepatocellularcarcinoma.

The results of our study together with other reports [18]suggest that the decrease of claudin-1 expression in coloniccarcinoma of higher grade due to its down-regulation mightcontribute to the more aggressive behavior of the tumors. This

could be explained by the fact that the preservation of cellpolarity and paracellular flux by many claudin molecules sug-gests that these proteins contribute to tumor suppressive func-

tions in epithelial neoplasia [43]. In addition, changes in or lossof expression of tight junction proteins such as claudins canlead to cellular disorientation and detachment, which are com-

monly seen in neoplasia [44].In the present work, a statistically significant inverse corre-

lation was found between claudin-1 expression and depth of

tumor invasion. This is in accordance with the findings re-ported by Tokes et al. [36] who reported a decreased clau-din-1 expression in invasive breast carcinoma.

130 E. Abdelzaher et al.

This phenomenon was explained by Chao et al. [45] who

studied claudin-1 expression in lung carcinoma. They reportedthat overexpression of claudin-1 suppressed cancer cell migra-tion, invasion and metastasis. They thus concluded that clau-din-1 is a cancer invasion and metastasis suppressor.

In contrast to our findings, Kingusa et al. [46], Leotlela et al.[47] and others [35,38] reported overexpression of claudin-1 ininvasive tumor cells. They explained their findings by the fact

that claudins are able to interact with membrane-type matrixmetalloproteinases (MT-MMPs) and thus promote MMPactivity. They thus concluded that claudin overexpression

seems to be an early event in carcinogenesis at least for somecancer types.

Also, in this study, a statistically significant strong inverse cor-

relation was seen between claudin-1 expression and lymph nodeinvolvement. Ersoz et al. [48] reported similar results and con-cluded that claudin-1 expression in colonic carcinoma was signif-icantly decreased with lymph node metastasis. These findings

were also in agreement with those reported by Cha et al. [42].These findings were explained by the fact that a modifica-

tion in the tight junction function and structure may affect tu-

mor penetration, [49] and that loss of claudins and other tightjunction proteins account for the loss of cell adhesion which isan important step in the development of tumor invasion and

metastasis [50].On the other hand, several authors reported an increase in

claudin-1 expression in metastatic carcinomas [26,45]. The rea-son for the discrepancy observed in claudin expression among

the different tissues is unclear but may be related to tissue-specific differences in claudin function or even in the tissuemicroenvironmental factors [51]. The discrepancy could also

be explained by the differences in methodology and antibodiesused.

In the present study, claudin-1 expression was also corre-

lated with the tumor stage which is until now considered themost important prognostic factor in colonic carcinoma [52].A statistically significant strong inverse correlation was found

between claudin-1 expression and tumor stage. These resultsare in accordance with those reported by Matsuoka et al.[40] who stated that a reduced expression of claudin-1 was cor-related with advanced tumor stage.

Also, the results are in agreement with other recent studies[20,35,53]. These studies highlighted the role that claudin-1plays in invasion and metastasis and tumor progression. In

contrast, Grone et al. [54] showed no clear correlation betweenclaudin-1 expression and tumor stage.

In the present work, regression analysis showed that clau-

din-1 decreased expression was a significant independent pre-dictor for tumor grade, stage and lymph node status.Decreased claudin-1 expression significantly predicts that the

tumor is of a high grade, high stage, and is associated withlymph node involvement.

An additional statistical test was done to delineate a cut offpoint that significantly discriminates between low and high

stage tumors as well as predicts lymph node status. A cutoffpoint was determined at a final score of �5.

At the cutoff point of �5, claudin-1 showed a sensitivity of

88.24% and a specificity of 81.25% for the prediction of tumorstage as well as a sensitivity of 73.33% and a specificity of82.86% for the prediction of lymph node status. The predictive

power was good.

Accordingly, claudin-1 decreased expression with a final

score of 6�5 significantly predicts that the tumor is of highstage (III or IV) and is associated with lymph node metastasis.On the other hand, if claudin-1 final score is > �5, it signifi-cantly predicts that the tumor is of low stage (I or II) and is

associated with negative lymph nodes.Thus, the level of claudin-1 expression could predict the

stage and lymph node status of an individual patient. How-

ever, these results need to be supported by further studies ona larger scale of patients.

From the present work it can be concluded that, claudin-1

is consistently expressed in non-neoplastic colonic mucosawhich emphasizes its important role in maintaining the normalfunction and morphology of colonocytes. Claudin-1 may be

involved in the process of colonic tumorigenesis. Claudin-1 de-creased expression contributes to the progression and dediffer-entiation of colonic carcinoma and it may underlie the invasiveand metastatic potential of colonic carcinoma. Immunohisto-

chemical claudin-1 expression could be used as a predictor ofcolonic carcinoma stage as well as lymph node status with ahigh sensitivity and specificity.

References

[1] Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics,

2002. CA Cancer J Clin 2005;55(2):74.

[2] Parkin D, Ferlay J, Cherif HM, Sitas F, Thomas J, Wabinga H,

et al. Cancer in Africa: epidemiology and prevention. Lyon:

IARC Press International Agency for Research on Cancer

(IARC); 2003 [277–281].

[3] El-Bolkainy N, Nouh M, El-Bolkainy T. Topographic

pathology of cancer. 3rd ed. Cairo: Egyptian National Cancer

Institiute; 2005, p. 83.

[4] Arnold CN, Goel A, Blum HE, Richard Boland C. Molecular

pathogenesis of colonic cancer. Cancer 2005;104(10):2035–47.

[5] Rowley PT. Inherited susceptibility to colonic cancer. Annu Rev

Med 2005;56:539–54.

[6] KahlenbergMS, Sullivan JM,WitmerDD,PetrelliNJ.Molecular

prognostics in colonic cancer. Surg Oncol 2003;12(3):173–86.

[7] Walther A, Johnstone E, Swanton C, Midgley R, Tomlinson I,

Kerr D. Genetic prognostic and predictive markers in colonic

cancer. Nat Rev Cancer 2011;11(4):309.

[8] Kim MS, Lee J, Sidransky D. DNA methylation markers in

colonic cancer. Cancer Metast Rev 2010;29(1):181–206.

[9] Andreyev J, Cunningham D. Markers, markers everywhere.

Prognosis in colonic cancer-time for a new approach. J Clin

Oncol 2001;19(2):286.

[10] Gonzalez-Mariscal L, Lechuga S, Garay E. Role of tight

junctions in cell proliferation and cancer. Prog Histochem

Cytochem 2007;42(1):1–57.

[11] Alrov J. Ultrastructure of canine urinary bladder carcinoma.

Vet Path Online 1979;16(6):693.

[12] Martinez-Palomo A. Ultrastructural modifications of

intercellular junctions in some epithelial tumors. Lab Invest

1970;22(6):605.

[13] Robenek H, Schopper C, Fasske E, Fetting R, Themann H.

Structure and function of the junctional complement of

spontaneous and transplanted murine mammary carcinomas. J

Submicrosc Cytol 1981;13(3):347.

[14] Swift J, Mukherjee T, Rowland R. Intercellular junctions in

hepatocellular carcinoma. J Submicrosc Cytol 1983;15(3):799.

[15] Resnick M, Konkin T, Routhier J, Sabo E, Pricolo V. Claudin-1

is a strong prognostic indicator in stage II colonic cancer: a

tissue microarray study. Mod Pathol 2004;18(4):511–8.

Predictive value of immunohistochemical expression of claudin-1 in colonic carcinoma 131

[16] Tsukita S, Furuse M, Itoh M. Multifunctional strands in tight

junctions. Nat Rev Mol Cell Biol 2001;2(4):285–93.

[17] Morin PJ. Claudin proteins in human cancer: promising new

targets for diagnosis and therapy. Cancer Res 2005;65(21):9603.

[18] Soini Y. Expression of claudins 1, 2, 3, 4, 5 and 7 in various

types of tumours. Histopathology 2005;46(5):551–60.

[19] Sheehan GM, Kallakury BVS, Sheehan CE, Fisher HAG,

Kaufman Jr RP, Ross JS. Loss of claudins-1 and -7 and

expression of claudins-3 and -4 correlate with prognostic variables

in prostatic adenocarcinomas. Hum Pathol 2007;38(4):564–9.

[20] Resnick MB, Konkin T, Routhier J, Sabo E, Pricolo VE.

Claudin-1 is a strong prognostic indicator in stage II colonic

cancer: a tissue microarray study. Mod Pathol 2005;18(4):511–8.

[21] Forster C. Tight junctions and the modulation of barrier

function in disease. Histochem Cell Biol 2008;130(1):55–70.

[22] Higashi Y, Suzuki S, Sakaguchi T, Nakamura T, Baba S, Reinecker

HC, et al. Loss of claudin-1 expression correlates with malignancy

of hepatocellular carcinoma. J Surg Res 2007;139(1):68–76.

[23] Akasaka H, Sato F, Morohashi S, Wu Y, Liu Y, Kondo J, et al.

Anti-apoptotic effect of claudin-1 in tamoxifen-treated human

breast cancer MCF-7 cells. BMC Cancer 2010;10:548.

[24] HuoQ,KinugasaT,WangL,HuangJ,ZhaoJ, ShibaguchiH, et al..

Claudin-1 protein is a major factor involved in the tumorigenesis of

colonic cancer. Anticancer Res 2009;29(3):851–7.

[25] Chang TL, Ito K, Ko TK, Liu Q, Salto-Tellez M, Yeoh KG,

et al. Claudin-1 has tumor suppressive activity and is a direct

target of RUNX3 in gastric epithelial cells. Gastroenterology

2010;138(1):255–65.

[26] Dhawan P, Singh AB, Deane NG, No Y, Shiou SR, Schmidt C,

et al. Claudin-1 regulates cellular transformation and metastatic

behavior in colon cancer. J Clin Invest 2005;115(7):1765–76.

[27] Stanley R, Hamilton LA. Pathology and genetics of tumours of

the digestive system. Lyon: IARC Press International Agency

for Research on Cancer (IARC), 2000;105–119.

[28] Green F, Balch C, Fleming I, Fritz A, Haller D, Morrow M,

et al. AJCC cancer staging handbook. 6th edition. New

York: Springer; 2002, p. 176–186.

[29] Cunningham D. Colonic cancer reply. Lancet

2010;376(9738):331–2.

[30] Rudmik L, Magliocco A. Molecular mechanisms of hepatic

metastasis in colonic cancer. J Surg Oncol 2005;92(4):347–59.

[31] Compton C, Fenoglio Preiser CM, Pettigrew N, Fielding LP.

American joint committee on cancer prognostic factors

consensus conference. Cancer 2000;88(7):1739–57.

[32] Baba Y, Nosho K, Shima K, Hayashi M, Meyerhardt JA, Chan

AT, et al. Phosphorylated AKT expression is associated with

PIK3CA mutation, low stage, and favorable outcome in 717

colonic cancers. Cancer 2011;117(7):1399–408.

[33] Singh AB, Sharma A, Dhawan P. Claudin family of proteins

and cancer: an overview. J Oncol 2010;2010:11.

[34] Hahn-Stromberg V, Edvardsson H, Bodin L, Franzen L.

Claudin 1 and claudin 7 gene polymorphisms and protein

derangement are unrelated to the growth pattern and tumor

volume of colon carcinoma. Int J Biomed Sci 2010;6(2):96–102.

[35] Miwa N, Furuse M, Tsukita S, Niikawa N, Nakamura Y,

Furukawa Y. Involvement of claudin-1 in the beta-catenin Tcf

signaling pathway and its frequent upregulation in human

colonic cancers. Oncol Res Feat Preclin Clin Cancer Therap

2001;11(12):469–76.

[36] Tokes AM, Kulka J, Paku S, Szik A, Paska C, Novak PK, et al.

Claudin-1, -3 and -4 proteins and mRNA expression in benign

and malignant breast lesions: a research study. Breast Cancer

Res 2005;7(2):296–305.

[37] Mullin JM. Potential interplay between luminal growth factors

and increased tight junction permeability in epithelial

carcinogenesis. J Exp Zool 1997;279(5):484–9.

[38] de Oliveira SS, de Oliveira IM, De Souza W, Morgado-Dıaz JA.

Claudins upregulation in human colonic cancer. FEBS Lett

2005;579(27):6179.

[39] Mitic L, Anderson J. Molecular architecture of tight junctions.

Annu Rev Physiol 1998;60(1):121–42.

[40] Matsuoka T, Mitomi H, Fukui N, Kanazawa H, Saito T,

Hayashi T, et al. Cluster analysis of claudin-1 and -4, E-

cadherin, and beta-catenin expression in colonic cancers. J Surg

Oncol 2011;103(7):674–86.

[41] Tzelepi VN, Tsamandas AC, Vlotinou HD, Vagianos CE, Scopa

CD. Tight junctions in thyroid carcinogenesis: diverse

expression of claudin-1, claudin-4, claudin-7 and occludin in

thyroid neoplasms. Mod Pathol 2008;21(1):22–30.

[42] Cha EJ, Jang KY, Choi HN, Kim KR, Choi KH,

Hwang SH, et al. Expression of claudin-1, beta-catenin and E-

cadherin in adenocarcinoma of the colon. Korean J Med

2008;74(2):146–53.

[43] Jakab C, Rusvai M, Galfi P, Szabo Z, Szabara A. Expression of

claudin-1, -3, -4, -5 and -7 proteins in low grade colonic

carcinoma of canines. Histol Histopathol 2010;25(1):55–62.

[44] French AD, Fiori JL, Camilli TC, Leotlela PD, O’Connell MP,

Frank BP, et al. PKC and PKA phosphorylation affect the

subcellular localization of claudin-1 in melanoma cells. Int J

Med Sci 2009;6(2):93–101.

[45] Chao YC, Pan SH, Yang SC, Yu SL, Che TF, Lin CW, et al.

Claudin-1 is a metastasis suppressor and correlates with clinical

outcome in lung adenocarcinoma. Am J Respir Crit Care Med

2008;179(2):123–33.

[46] Kinugasa T, Huo Q, Higashi D, Shibaguchi H, Kuroki M,

Tanaka T, et al. Selective up-regulation of claudin-1 and

claudin-2 in colonic cancer. Anticancer Res 2007;27(6A):3729.

[47] Leotlela PD, Wade MS, Duray PH, Rhode MJ, Brown HF,

Rosenthal DT, et al. Claudin-1 overexpression in melanoma is

regulated by PKC and contributes to melanoma cell motility.

Oncogene 2007;26(26):3846–56.

[48] Ersoz S, Mungan S, Cobanoglu U, Turgutalp H, Ozoran Y.

Prognostic importance of claudin-1 and claudin-4 expression in

colon carcinomas. Pathol Res Pract 2011;207(5):285–9.

[49] Jiang WG, Martin T, Llaffafian I, Mansel RE. Tight junctions, a

critical structure in the control of cancer invasion and

metastasis. Cancer Metast Mol Cell Mech Clin Interv

2004:195–213.

[50] Swisshelm K, Macek R, Kubbies M. Role of claudins in

tumorigenesis. Adv Drug Deliv Rev 2005;57(6):919–28.

[51] Oliveira SS, Morgado-Dıaz JA. Claudins: multifunctional

players in epithelial tight junctions and their role in cancer.

Cell Mol Life Sci 2007;64(1):17–28.

[52] Kehoe J, Khatri VP. Staging and prognosis of colon cancer.

Surg Oncol Clin N Am 2006;15(1):129.

[53] Hoevel T, Macek R, Swisshelm K, Kubbies M. Reexpression of

the TJ protein CLDN1 induces apoptosis in breast tumor

spheroids. Int J Cancer 2004;108(3):374–83.

[54] Grone J, Weber B, Staub E, Heinze M, Klaman I, Pilarsky C,

et al.. Differential expression of genes encoding tight junction

proteins in colonic cancer: frequent dysregulation of claudin-1,-8

and-12. Int J Colon Dis 2007;22(6):651–9.